Using Genetics to Inform Interventions Related to Sodium and Potassium in Hypertension

Abstract

Background: Hypertension is a key risk factor for major adverse cardiovascular events but remains difficult to treat in many individuals. Dietary interventions are an effective approach to lower blood pressure (BP) but are not equally effective across all individuals. BP is heritable, and genetics may be a useful tool to overcome treatment response heterogeneity. We investigated whether the genetics of BP could be used to identify individuals with hypertension who may receive a particular benefit from lowering sodium intake and boosting potassium levels.

Methods: In this observational genetic study, we leveraged cross-sectional data from up to 296 475 genotyped individuals drawn from the UK Biobank cohort for whom BP and urinary electrolytes (sodium and potassium), biomarkers of sodium and potassium intake, were measured. Biologically directed genetic scores for BP were constructed specifically among pathways related to sodium and potassium biology (pharmagenic enrichment scores), as well as unannotated genome-wide scores (conventional polygenic scores). We then tested whether there was a gene-by-environment interaction between urinary electrolytes and these genetic scores on BP.

Results: Genetic risk and urinary electrolytes both independently correlated with BP. However, urinary sodium was associated with a larger BP increase among individuals with higher genetic risk in sodium- and potassium-related pathways than in those with comparatively lower genetic risk. For example, each SD in urinary sodium was associated with a 1.47-mm Hg increase in systolic BP for those in the top 10% of the distribution of genetic risk in sodium and potassium transport pathways versus a 0.97-mm Hg systolic BP increase in the lowest 10% (P=1.95×10^-3). This interaction with urinary sodium remained when considering estimated glomerular filtration rate and indexing sodium to urinary creatinine. There was no strong evidence of an interaction between urinary sodium and a standard genome-wide polygenic score of BP.

Conclusions: The data suggest that genetic risk in sodium and potassium pathways could be used in a precision medicine model to direct interventions more specifically in the management of hypertension. Intervention studies are warranted.

Keywords: genetics; hypertension; potassium; precision medicine; sodium.

Figure 1.

Development of pharmagenic enrichment scores to inform precision interventions related to sodium and potassium in blood pressure. A, Condensed overview of the methodology applied in this study to implement the pharmagenic enrichment score (PES) and evaluate its properties. B, Venn diagrams depicting the overlap in unique genes between different collated ontologies related to sodium and potassium biology. From left to right: sodium transport genes versus potassium transport genes; renal sodium excretion genes versus renal potassium genes; intersection of the 4 previous gene sets. BP indicates blood pressure; HCS, Hunter Community Study; and UKBB, UK Biobank.

Figure 2.

Association of urinary sodium with systolic blood pressure with increasing sodium/potassium transport pharmagenic enrichment score. A, The systolic blood pressure (SBP) effect size (mm Hg) per SD of scaled urinary sodium (SD=1) in each percentile of the sodium and potassium transport pharmagenic enrichment score (PES). The per-percentile estimates were grouped as low (1st–20th percentile), moderate (20th–80th percentile), or high (80th–top percentile). Box whisker plots were overlaid on the effect sizes in each group, denoting the median (± interquartile range) urinary sodium (UNa⁺) effect size for each group. The dotted line represents the mean SBP effect size across all percentiles. B, Analogous to A, but the percentiles instead represent percentiles of genome-wide SBP polygenic score (PGS). C, Smoothed trendlines of the effect size of UNa⁺ on SBP in each increasing percentile of the PES. The first panel was smoothed using a generalized linear model (GLM), the second panel adds a quadratic term, and the third panel uses local polynomial regression (locally estimated scatterplot smoothing [LOESS]). All panels demonstrate the larger effect size association of urinary sodium with SBP at elevated PES.